Examples with CoreLabel edu.stanford.nlp.ling.CoreLabel used on opensource projects

Example 41 with CoreLabel

use of edu.stanford.nlp.ling.CoreLabel in project CoreNLP by stanfordnlp.

the class CorefMentionFinder method safeHead.

private Tree safeHead(Tree top, int endIndex) {
    // The trees passed in do not have the CoordinationTransformer
    // applied, but that just means the SemanticHeadFinder results are
    // slightly worse.
    Tree head = top.headTerminal(headFinder);
    // One obscure failure case is that the added period becomes the head. Disallow this.
    if (head != null) {
        Integer headIndexInteger = ((CoreLabel) head.label()).get(CoreAnnotations.IndexAnnotation.class);
        if (headIndexInteger != null) {
            int headIndex = headIndexInteger - 1;
            if (headIndex < endIndex) {
                return head;
            }
        }
    }
    // if no head found return the right-most leaf
    List<Tree> leaves = top.getLeaves();
    int candidate = leaves.size() - 1;
    while (candidate >= 0) {
        head = leaves.get(candidate);
        Integer headIndexInteger = ((CoreLabel) head.label()).get(CoreAnnotations.IndexAnnotation.class);
        if (headIndexInteger != null) {
            int headIndex = headIndexInteger - 1;
            if (headIndex < endIndex) {
                return head;
            }
        }
        candidate--;
    }
    // fallback: return top
    return top;
}

Example 42 with CoreLabel

use of edu.stanford.nlp.ling.CoreLabel in project CoreNLP by stanfordnlp.

the class DependencyCorefMentionFinder method extractMentionForHeadword.

private void extractMentionForHeadword(IndexedWord headword, SemanticGraph dep, CoreMap s, List<Mention> mentions, Set<IntPair> mentionSpanSet, Set<IntPair> namedEntitySpanSet) {
    List<CoreLabel> sent = s.get(CoreAnnotations.TokensAnnotation.class);
    SemanticGraph basic = s.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
    SemanticGraph enhanced = s.get(SemanticGraphCoreAnnotations.EnhancedDependenciesAnnotation.class);
    if (enhanced == null) {
        enhanced = s.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
    }
    // pronoun
    if (headword.tag().startsWith("PRP")) {
        extractPronounForHeadword(headword, dep, s, mentions, mentionSpanSet, namedEntitySpanSet);
        return;
    }
    // add NP mention
    IntPair npSpan = getNPSpan(headword, dep, sent);
    int beginIdx = npSpan.get(0);
    int endIdx = npSpan.get(1) + 1;
    // try not to have span that ends with ,
    if (",".equals(sent.get(endIdx - 1).word())) {
        endIdx--;
    }
    // try to remove first IN.
    if ("IN".equals(sent.get(beginIdx).tag())) {
        beginIdx++;
    }
    addMention(beginIdx, endIdx, headword, mentions, mentionSpanSet, namedEntitySpanSet, sent, basic, enhanced);
    //
    // extract the first element in conjunction (A and B -> extract A here "A and B", "B" will be extracted above)
    //
    // to make sure we find the first conjunction
    Set<IndexedWord> conjChildren = dep.getChildrenWithReln(headword, UniversalEnglishGrammaticalRelations.CONJUNCT);
    if (conjChildren.size() > 0) {
        IndexedWord conjChild = dep.getChildWithReln(headword, UniversalEnglishGrammaticalRelations.CONJUNCT);
        for (IndexedWord c : conjChildren) {
            if (c.index() < conjChild.index())
                conjChild = c;
        }
        IndexedWord left = SemanticGraphUtils.leftMostChildVertice(conjChild, dep);
        for (int endIdxFirstElement = left.index() - 1; endIdxFirstElement > beginIdx; endIdxFirstElement--) {
            if (!sent.get(endIdxFirstElement - 1).tag().matches("CC|,")) {
                if (headword.index() - 1 < endIdxFirstElement) {
                    addMention(beginIdx, endIdxFirstElement, headword, mentions, mentionSpanSet, namedEntitySpanSet, sent, basic, enhanced);
                }
                break;
            }
        }
    }
}

Example 43 with CoreLabel

use of edu.stanford.nlp.ling.CoreLabel in project CoreNLP by stanfordnlp.

the class DependencyCorefMentionFinder method findHeadInDependency.

// TODO: still errors in head finder
public static void findHeadInDependency(CoreMap s, Mention m) {
    List<CoreLabel> sent = s.get(CoreAnnotations.TokensAnnotation.class);
    SemanticGraph basicDep = s.get(BasicDependenciesAnnotation.class);
    if (m.headWord == null) {
        // when there's punctuation, no node found in the dependency tree
        int curIdx;
        IndexedWord cur = null;
        for (curIdx = m.endIndex - 1; curIdx >= m.startIndex; curIdx--) {
            if ((cur = basicDep.getNodeByIndexSafe(curIdx + 1)) != null)
                break;
        }
        if (cur == null)
            curIdx = m.endIndex - 1;
        while (cur != null) {
            IndexedWord p = basicDep.getParent(cur);
            if (p == null || p.index() - 1 < m.startIndex || p.index() - 1 >= m.endIndex)
                break;
            curIdx = p.index() - 1;
            cur = basicDep.getNodeByIndexSafe(curIdx + 1);
        }
        //      for(IndexedWord p : basicDep.getPathToRoot(basicDep.getNodeByIndex(curIdx+1))) {
        //        if(p.index()-1 < m.startIndex || p.index()-1 >= m.endIndex) {
        //          break;
        //        }
        //        curIdx = p.index()-1;
        //      }
        m.headIndex = curIdx;
        m.headWord = sent.get(m.headIndex);
        m.headString = m.headWord.word().toLowerCase(Locale.ENGLISH);
    }
}

Example 44 with CoreLabel

use of edu.stanford.nlp.ling.CoreLabel in project CoreNLP by stanfordnlp.

the class HybridCorefMentionFinder method extractNPorPRP.

private static void extractNPorPRP(CoreMap s, List<Mention> mentions, Set<IntPair> mentionSpanSet, Set<IntPair> namedEntitySpanSet) {
    List<CoreLabel> sent = s.get(CoreAnnotations.TokensAnnotation.class);
    Tree tree = s.get(TreeCoreAnnotations.TreeAnnotation.class);
    tree.indexLeaves();
    SemanticGraph basicDependency = s.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
    SemanticGraph enhancedDependency = s.get(SemanticGraphCoreAnnotations.EnhancedDependenciesAnnotation.class);
    if (enhancedDependency == null) {
        enhancedDependency = s.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
    }
    TregexPattern tgrepPattern = npOrPrpMentionPattern;
    TregexMatcher matcher = tgrepPattern.matcher(tree);
    while (matcher.find()) {
        Tree t = matcher.getMatch();
        List<Tree> mLeaves = t.getLeaves();
        int beginIdx = ((CoreLabel) mLeaves.get(0).label()).get(CoreAnnotations.IndexAnnotation.class) - 1;
        int endIdx = ((CoreLabel) mLeaves.get(mLeaves.size() - 1).label()).get(CoreAnnotations.IndexAnnotation.class);
        // try not to have span that ends with ,
        if (",".equals(sent.get(endIdx - 1).word())) {
            endIdx--;
        }
        IntPair mSpan = new IntPair(beginIdx, endIdx);
        //      if(!mentionSpanSet.contains(mSpan) && (!insideNE(mSpan, namedEntitySpanSet)) ) {
        if (!mentionSpanSet.contains(mSpan) && (!insideNE(mSpan, namedEntitySpanSet) || t.value().startsWith("PRP"))) {
            int dummyMentionId = -1;
            Mention m = new Mention(dummyMentionId, beginIdx, endIdx, sent, basicDependency, enhancedDependency, new ArrayList<>(sent.subList(beginIdx, endIdx)), t);
            mentions.add(m);
            mentionSpanSet.add(mSpan);
            if (m.originalSpan.size() > 1) {
                boolean isNE = true;
                for (CoreLabel cl : m.originalSpan) {
                    if (!cl.tag().startsWith("NNP"))
                        isNE = false;
                }
                if (isNE) {
                    namedEntitySpanSet.add(mSpan);
                }
            }
        }
    }
}

Example 45 with CoreLabel

use of edu.stanford.nlp.ling.CoreLabel in project CoreNLP by stanfordnlp.

the class HybridCorefMentionFinder method extractNamedEntityMentions.

protected static void extractNamedEntityMentions(CoreMap s, List<Mention> mentions, Set<IntPair> mentionSpanSet, Set<IntPair> namedEntitySpanSet) {
    List<CoreLabel> sent = s.get(CoreAnnotations.TokensAnnotation.class);
    SemanticGraph basicDependency = s.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
    SemanticGraph enhancedDependency = s.get(SemanticGraphCoreAnnotations.EnhancedDependenciesAnnotation.class);
    if (enhancedDependency == null) {
        enhancedDependency = s.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
    }
    String preNE = "O";
    int beginIndex = -1;
    for (CoreLabel w : sent) {
        String nerString = w.ner();
        if (!nerString.equals(preNE)) {
            int endIndex = w.get(CoreAnnotations.IndexAnnotation.class) - 1;
            if (!preNE.matches("O")) {
                if (w.get(CoreAnnotations.TextAnnotation.class).equals("'s") && w.tag().equals("POS")) {
                    endIndex++;
                }
                IntPair mSpan = new IntPair(beginIndex, endIndex);
                // attached to the previous NER by the earlier heuristic
                if (beginIndex < endIndex && !mentionSpanSet.contains(mSpan)) {
                    int dummyMentionId = -1;
                    Mention m = new Mention(dummyMentionId, beginIndex, endIndex, sent, basicDependency, enhancedDependency, new ArrayList<>(sent.subList(beginIndex, endIndex)));
                    mentions.add(m);
                    mentionSpanSet.add(mSpan);
                    namedEntitySpanSet.add(mSpan);
                }
            }
            beginIndex = endIndex;
            preNE = nerString;
        }
    }
    // NE at the end of sentence
    if (!preNE.matches("O")) {
        IntPair mSpan = new IntPair(beginIndex, sent.size());
        if (!mentionSpanSet.contains(mSpan)) {
            int dummyMentionId = -1;
            Mention m = new Mention(dummyMentionId, beginIndex, sent.size(), sent, basicDependency, enhancedDependency, new ArrayList<>(sent.subList(beginIndex, sent.size())));
            mentions.add(m);
            mentionSpanSet.add(mSpan);
            namedEntitySpanSet.add(mSpan);
        }
    }
}